10. Assessment of the Northern Rockfish Stock in the Gulf of Alaska

Executive Summary

We use a statistical age-structured model as the primary assessment tool for GOA northern rockfish which qualifies as a Tier 3 stock. This assessment consists of a population model, which uses survey and fishery data to generate a historical time series of population estimates, and a projection model, which uses results from the population model to predict future population estimates and recommended harvest levels. The data sets used in this assessment include total catch biomass, fishery age and size compositions, trawl survey abundance estimates, and trawl survey age compositions. For Gulf of Alaska northern rockfish in 2022, we present a full assessment with updated assessment and projection model results to recommend harvest levels for the next two years.

Summary of Changes in Assessment Inputs

Changes to input data: Relative to the last full assessment the following substantive changes have been made to assessment inputs:

• include survey biomass estimates for 2021,
• update survey age compositions with 2021 data,
• update fishery age compositions with 2020 data,
• update final catch values for 2020 and 2021, and use preliminary catch for 2022.

The survey biomass estimate is based upon the Groundfish Assessment Program’s Vector Autoregressive Spatio-temporal (VAST) model for the GOA through 2021.
Survey data from the 1980s were excluded from this assessment. In the last few assessments survey biomass from 1984 and 1987 have been included in the survey biomass estimate (though not in the compositional data), however those surveys used different vessels and gear and are not directly comparable to survey data from 1990+. Therefore they have been excluded from this assessment going forward.

Changes in assessment methodology: The following model change is recommended in the current assessment: extend the length plus group from 38 cm to 45 cm.

Summary of Results

A suite of incremental models were run to investigate the effects of removing 1980s survey data from the assessment, and increasing the length plus group.

The author’s preferred model is m22.1, which is the 2020 model with updated data through 2022, and an increased length plus group that uses a VAST model-based index of survey abundance with GAP default settings. This model generally produces good visual fits to the data and biologically reasonable patterns of recruitment, abundance, and selectivity, and relatively low retrospective Mohn’s rho value.

The m22.1 projected age 2+ total biomass for 2023 is 95,452 t. The recommended ABC for 2023 is 4,965 t, the maximum allowable ABC under Tier 3a. This ABC is a -4% increase compared to the 2022 ABC of 5,147 t and a 1% increase from the projected 2023 ABC from last year. The 2023 GOA-wide OFL for northern rockfish is 5,927 t.

The stock is not being subject to overfishing, is not currently overfished, nor is it approaching a condition of being overfished.

Reference values for northern rockfish are summarized in the following table:

Area Allocation of Harvest

The following table shows the recommended ABC apportionment for 2022 and 2023. Apportionment is based on the random effects model developed by Plan Team survey averaging working group, which was fit to area-specific design-based biomass indices through 2021 from the bottom trawl survey.

Amendment 41 prohibited trawling in the Eastern area east of 140° W longitude. The ratio of biomass still obtainable in the W. Yakutat area (between 147° W and 140° W) is 0.75. This results in the following apportionment to the W. Yakutat area:

Responses to SSC and Plan Team Comments on Assessments in General

“The SSC requests that all authors fill out the risk table in 2019…” (SSC December 2018)

A risk table has been included in this full assessment.

“The Team recommends all GOA authors evaluate any bottom trawl survey information used in their assessment prior to 1990 including the 1984 and 1987 surveys and conduct sensitivity analyses to evaluate their usefulness to the assessment. This may apply for Aleutian Islands surveys but this was only raised during GOA assessment considerations.” (PT, November 2021)

In recent assessments survey biomass estimates from 1984 and 1987 (note that age and size compistion data were not included) have been included in the survey biomass estimate, however those surveys used different vessels and gear and are not directly comparable to survey data from 1990+. A suite of incremental models were run to investigate the effects of removing 1980s survey data from the assessment. Therefore they have been excluded from this assessment going forward.

Responses to SSC and Plan Team Comments Specific to this Assessment

The Team recommends evaluating how the definition of the length composition plus group, and alternative data-weighting methods, affect model performance.” (Plan Team, November 2015); “Finally, the SSC notes the increasing proportion of fish in the fishery length composition plus-group and looks forward to seeing the results of the ongoing investigations into alternative length composition bin structures. The SSC also agrees with the high priority placed on improving maturity-at-age information for northern rockfish.” (SSC, December 2018)

The authors have increased the length plus group by 5 cm and present finding as the author’s recommended model. Alternative data-weightings have been examined but are not presented in this assessent due to computational issues, these results will be presented in the next full assessment. Skip spawning has been observed for northern rockfish, however the spatial and temporal extent are unknown. A prelimiary examination of spawning biomass estimates with skip spawning is included in the appendix, though further examination is warranted.

Introduction

Biology and Distribution

Northern rockfish, Sebastes polyspinis, is a locally abundant and commercially valuable member of its genus in Alaskan waters. As implied by its common name, northern rockfish has one of the most northerly distributions among the 60+ species of Sebastes in the North Pacific Ocean. It ranges from extreme northern British Columbia around the northern Pacific Rim to eastern Kamchatka and the northern Kuril Islands and also north into the eastern Bering Sea (Allen and Smith 1988). Within this range, northern rockfish are most abundant in Alaska waters, from the western end of the Aleutian Islands to Portlock Bank in the central Gulf of Alaska (Clausen and Heifetz 2002).

Little is known about the life history of northern rockfish. Like other Sebastes species, northern rockfish are presumed to be ovoviviparous with internal fertilization. There have been no studies on fecundity of northern rockfish. Observations during research surveys in the Gulf of Alaska indicate that parturition (larval release) occurs in the spring and is completed by summer. Larval northern rockfish cannot be unequivocally identified to species at this time, even using genetic techniques, so information on larval distribution and length of the larval stage is unknown. The larvae metamorphose to a pelagic juvenile stage, but there is no information on when these juveniles become demersal.

Little information is available on the habitat of juvenile northern rockfish. Studies in the eastern Gulf of Alaska and Southeast Alaska using trawls and submersibles have indicated that several species of juvenile (< 20 cm) red rockfish (Sebastes spp.) associate with benthic nearshore living and non-living structure and appear to use the structure as a refuge (Carlson and Straty 1981; Krieger 1993). (Freese2003?) also identified juvenile (5 to 10 cm) red rockfish (Sebastes spp.) associated with sponges (primarily Aphrocallistes spp.) attached to boulders 50 km offshore in the GOA at 148 m depth over a substrate that was primarily a sand and silt mixture. Only boulders with sponges harbored juvenile rockfish, and the juvenile red rockfish appeared to be using the sponges as shelter (Freese2003?). Although these studies did not specifically observe northern rockfish, it is likely that juvenile northern rockfish also utilize similar habitats. Length frequencies of northern rockfish captured in NMFS bottom trawl surveys and observed in commercial fishery bottom trawl catches indicate that older juveniles (>20 cm) are found on the continental shelf, generally at locations inshore of the adult habitat (Pers. comm. Dave Clausen).

Northern rockfish are generally planktivorous. They eat mainly euphausiids and calanoid copepods in both the GOA and the Aleutian Islands (Yang 1993, 1996, 2000). There is no indication of a shift in diet over time or a difference in diet between the GOA and AI Yang (2000). In the Aleutian Islands, calanoid copepods were the most important food of smaller-sized northern rockfish (< 25 cm), while euphausiids were the main food of larger sized fish (> 25 cm) (Yang 1996). The largest size group also consumed myctophids and squids (Yang 2000). Arrow worms, hermit crabs, and shrimp have also been noted as prey items in much smaller quantities Yang (1996). Large offshore euphausiids are not directly associated with the bottom, but rather, are thought to be advected onshore near bottom at the upstream ends of underwater canyons where they become easy prey for planktivorous fishes Brodeur (2001)]. Predators of northern rockfish are not well documented, but likely include larger fish, such as Pacific halibut, that are known to prey on other rockfish species.

Trawl surveys and commercial fishing data indicate that the preferred habitat of adult northern rockfish in the Gulf of Alaska is relatively shallow rises or banks on the outer continental shelf at depths of about 75-150 m (Clausen and Heifetz 2002). The highest concentrations of northern rockfish from NMFS trawl survey catches appear to be associated with relatively rough (variously defined as hard, steep, rocky or uneven) bottom on these banks (Clausen and Heifetz 2002). Heifetz (2002) identified rockfish as among the most common commercial fish captured with gorgonian corals (primarily Callogorgia, Primnoa, Paragorgia, Fanellia, Thouarella, and Arthrogorgia) in NMFS trawl surveys of Gulf of Alaska and Aleutian waters. Krieger and Wing (2002) identified six rockfish species associated with gorgonian coral (Primnoa spp.) from a manned submersible in the eastern Gulf of Alaska. Research focusing on non-trawlable habitats found rockfish species often associate with biogenic structure (Du Preez and Tunnicliffe 2011; Laman et al. 2015). However, most of these studies did not specifically observe northern rockfish, and more research is required to determine if northern rockfish are associated with living structure, including corals, in the Gulf of Alaska, and the nature of those associations if they exist. Recent work on black rockfish (Sebastes melanops) has shown that larval survival may be higher from older female spawners (Berkeley et al. 2004). The black rockfish population has shown a distinct reduction in the proportion of older fish in recent fishery samples off the West Coast of North America, raising concerns if larval survival diminishes with spawner age. Bruin et al. (2004) examined Pacific ocean perch (S. alutus) and rougheye rockfish (S. aleutianus) for senescence in reproductive activity of older fish and found that oogenesis continues at advanced ages. Leaman (1991) showed that older individuals have slightly higher egg dry weight than their middle-aged counterparts. Some literature suggests that environmental factors may affect the condition of female rockfish that contributes to reproductive success (Hannah and Parker 2007; Rodgveller et al. 2012; Beyer et al. 2015). However, relationships on fecundity or larval survival at age have not yet been evaluated for northern rockfish or other rockfish in Alaska. Stock assessments for Alaska groundfish have assumed that the reproductive success of mature fish is independent of age.

Stock Structure

Gulf of Alaska northern rockfish grow significantly faster and reach a larger maximum length than Aleutian Islands northern rockfish (Clausen and Heifetz 2002). Also, Aleutian Islands northern rockfish are slightly older (maximum age-72) than GOA northern rockfish (maximum age-67), the difference in age could be due to sampling variability. There have been two studies on the genetic stock structure of northern rockfish. One study of northern rockfish provided no evidence for genetically distinct stock structure when comparing samples from near the western Aleutian Islands, the western GOA, and Kodiak Island (Gharrett et al. 2003). The results from that study were considered preliminary, and sample sizes were small. Consequently, the lack of evidence for stock structure did not necessarily confirm stock homogeneity. A more recent study did find spatial structure on a relatively small scale for northern rockfish sampled from several locations in the Aleutian Islands and Bering Sea (Gharrett et al. 2012).

Results of an analysis of localized depletion based on Leslie depletion estimators on targeted rockfish catches detected relatively few localized depletions for northern rockfish (Hanselman2007?). Several significant depletions occurred in the early 1990s for northern rockfish, but were not detected again by the depletion analysis. However, when fishery and survey CPUEs were plotted over time for a geographic block of high rockfish fishing intensity that contained the “Snakehead” area, the results indicated there were year-after-year drops in both fishery and survey CPUE for northern rockfish. The significance of these observations depends on the migratory and stock structure patterns of northern rockfish. If fine-scale stock structure is determined in northern rockfish, or if the area is essential to northern rockfish reproductive success, then these results would suggest that current apportionment of ABC may not be sufficient to protect northern rockfish from localized depletion. Provisions to guard against serial depletion in northern rockfish should be examined in the Gulf of Alaska rockfish rationalization plan. The extension of the fishing season that has been implemented may spread out the fishery in time and space and reduce the risk of localized serial depletion on the “Snakehead” and other relatively shallow (75 – 150 m) offshore banks on the outer continental shelf where northern rockfish are concentrated.

If there is relatively small scale stock structure (120 km) in Gulf of Alaska northern rockfish, then recovery from localized depletion, as indicated above for a region known as the “Snakehead,” could be slow. Analysis of otolith microchemistry may provide a useful tool, in addition to genetic analysis, for identifying small scale (120 km) stock structure of northern rockfish relative to their overall range. Berkeley et al. (2004) suggests that, in addition to the maintenance of age structure, the maintenance of spatial distribution of recruitment is essential for long-term sustainability of exploited rockfish populations. In particular, Berkeley et al. (2004) outline Hedgecock’s “sweepstakes hypothesis” to explain small-scale genetic heterogeneity observed in some widely distributed marine populations. According to Berkeley et al. (2004), “most spawners fail to produce surviving offspring because their reproductive activity is not matched in space and time to favorable oceanographic conditions for larval survival during a given season. As a result of this mismatch the surviving year class of new recruits is produced by only a small minority of adults that spawned within those restricted temporal and spatial oceanographic windows that offered good conditions for larval survival and subsequent recruitment”. However, (Miller2004?) found limited larval dispersal (120 km) in black rockfish off the Pacific coast with an analysis of otolith microchemistry. In particular, these results suggest that black rockfish exhibit some degree of stock structure at very small scales (120 km) relative to their overall range. Localized genetic stocks of Pacific ocean perch have also been found in northern B.C. (Withler et al. 2001), and (Kamin et al. 2013) concluded that fine-scale genetic heterogeneity for Pacific ocean perch in Alaska was not the influence of a sweepstakes effect. Limited larval dispersal contradicts Hedgecock’s hypothesis and suggests that genetic heterogeneity in rockfish may be the result of stock structure rather than the result of the sweepstakes hypothesis.

Fishery

Description of the Directed Fishery

In the Gulf of Alaska, northern rockfish are generally caught with bottom trawls identical to those used in the Pacific ocean perch (S. alutus) fishery. Many of these nets are equipped with so-called “tire gear” in which automobile tires are attached to the footrope to facilitate towing over rough substrates. Most of the catch has been taken during July, as the directed rockfish trawl fishery in the Gulf of Alaska has traditionally opened around July 1. Rockfish trawlers usually direct their efforts first toward Pacific ocean perch because of its higher value relative to other rockfish species. After the TAC for Pacific ocean perch has been reached and NMFS closes directed fishing for this species, trawlers switch and target northern rockfish. With implementation of the Central Gulf Rockfish Pilot Project in 2007, catches have been spread out more throughout the year.

Historically, bottom trawls have accounted for nearly all the commercial harvest of northern rockfish in the Gulf of Alaska. In the years 1990-98, bottom trawls took over 99% of the catch (Clausen and Heifetz 2002). Before 1996, most of the slope rockfish trawl catch (>90%) was taken by large factory-trawlers that processed the fish at sea. A significant change occurred in 1996, however, when smaller shore-based trawlers began taking a sizable portion of the catch in the Central Gulf for delivery to processing plants in Kodiak. Factory trawlers continued to take nearly all the northern rockfish catch in the Western area during this period.

A study of the northern rockfish fishery for the period 1990-98 showed that 89% of northern rockfish catch was taken from just five relatively small fishing grounds: Portlock Bank, Albatross Bank, an unnamed bank south of Kodiak Island that fishermen commonly refer to as the “Snakehead”, Shumagin Bank, and Davidson Bank (Clausen and Heifetz 2002). The Snakehead accounted for 46% of the northern rockfish catch during these years. All of these grounds can be characterized as relatively shallow (75–150 m) offshore banks on the outer continental shelf.

Data from the observer program for 1990-98 indicated that 82% of the northern rockfish catch during that period came from directed fishing for northern rockfish and 18% was taken as incidental catch in fisheries for other species (Clausen and Heifetz 2002).

Catch Patterns

Total commercial catch (t) of northern rockfish in the GOA for the years 1961-2018 is summarized by foreign, joint venture, and domestic fisheries (Table 10-?? and Figure 10-3). Catches of GOA northern rockfish during the years 1961-1976 were estimated as 5% of the foreign GOA Pacific ocean perch catch in the same years. A Pacific ocean perch trawl fishery by the U.S.S.R. and Japan began in the Gulf of Alaska in the early 1960’s. This fishery developed rapidly with massive efforts by the Soviet and Japanese fleets. Catches peaked in 1965 when a total of nearly 350,000 metric tons (t) were caught, but declined to 45,500 t by 1976. Some northern rockfish were likely taken in this fishery, but there are no available summaries of northern rockfish catches for this period. Foreign catches of all rockfish were often reported simply as “Pacific ocean perch” with no attempt to differentiate species. The only detailed analysis of bycatch in slope rockfish fisheries of the Gulf of Alaska is that of Ackley and Heifetz (2001) who examined data from the observer program for the years 1993-95. Consequently, our best estimate of northern rockfish catch from 1961-1976 comes from analysis of the ratio of northern rockfish catch to Pacific ocean perch catch in the years 1993-1995. For hauls targeting on Pacific ocean perch, northern rockfish composed 5% of the catch (Ackley and Heifetz 2001).

Catches of GOA northern rockfish during the years 1977-1983 were available from NMFS foreign and joint venture fisheries observer data. With the advent of a NMFS observer program aboard foreign fishing vessels in 1977, enough information on species composition of rockfish catches was collected so that estimates of the northern rockfish catch were made for 1977-83 from extrapolation of catch compositions from the foreign observer program (Clausen and Heifetz 2002). The relatively large catch estimates for the foreign fishery in 1982-83 are an indication that at least some directed fishing for northern rockfish probably occurred in those years. Joint venture catches of northern rockfish, however, appear to have been relatively modest.

Catches of GOA northern rockfish during the years 1984-1989 were estimated as 8% of the domestic slope rockfish catch during the same years. A completely domestic trawl fishery for rockfish in the Gulf of Alaska began in 1984 but a domestic observer program was not implemented until 1990. Domestic catches of GOA northern rockfish during the years 1984-1989 were estimated from the ratio of domestic northern rockfish catch to domestic slope rockfish catch (8%) reported by the 1990 NMFS observer program:

\[northern~rockfish~catch_i = \frac{northern~rockfish~catch_{1990}}{slope~rockfish~assemblage~catch_{1990}} \times slope~rockfish~assemblage~catch_i \]

Catches of GOA northern rockfish during the years 1990-1992 were estimated from extrapolation of catch compositions from the domestic observer program (Clausen and Heifetz 2002). Catch estimates of northern rockfish increased greatly from about 1,700 t in 1990 to nearly 7,800 t in 1992. The increases for 1991 and 1992 can be explained by the removal of Pacific ocean perch and shortraker/rougheye rockfish from the slope rockfish management group. As a result of this removal, relatively low TAC’s were adopted for these three species, and the rockfish fleet redirected more of its effort to northern rockfish in 1991 and 1992.

Catches of GOA northern rockfish during the years 1993-2022 were available directly from NMFS domestic fisheries observer data. Northern rockfish were removed from the slope rockfish assemblage and managed with an individual TAC beginning in 1993. As a consequence, directly reported catch for northern rockfish has been available since 1993. Catch of northern rockfish was reduced after the implementation of a northern specific TAC in 1993. Most of the catch since 1993 has been taken in the Central GOA, where the majority of the northern rockfish exploitable biomass is located. Gulfwide catches for the years 1993-2022 have ranged from 1,835 t to 5,966 t. Annual ABCs and TACs have been relatively consistent during this period and have varied between 3,000–6,000 t. In 2001, catch of northern rockfish was below TAC because the maximum allowable bycatch of Pacific halibut was reached in the central Gulf of Alaska for “deep water trawl species,” which includes northern rockfish. Catches of northern rockfish were near their TAC’s in 2003 – 2016, however in 2017 catch was 48% of the TAC and 2022 projected catch is likely to reach 53% of the TAC. Consultation with industry representatives suggested the low catch to TAC ratio in 2017 was largely driven by the fleet targeting alternative higher value species. Research catches of northern rockfish have been relatively small and are listed in Table 10a-1 in Appendix 10A.

Bycatch and Discards

The only detailed analysis of incidental catch in slope rockfish fisheries of the Gulf of Alaska is that of Ackley and Heifetz (2001) who examined data from the observer program for the years 1993-95. For hauls targeting on northern rockfish, the predominant incidental species were dusky rockfish, distantly followed by “other slope rockfish,” Pacific ocean perch, and arrowtooth flounder.

Total FMP groundfish catch estimates in the GOA rockfish fishery from 2018–2022 are shown in Table 10-3. As an average for the GOA rockfish fishery during 2018–2022, the largest non-rockfish bycatch groups are arrowtooth flounder, sablefish, atka mackerel and walleye pollock. Non-FMP species catch in the rockfish target fisheries is dominated by giant grenadier (Albatrossia pectoralis) and miscellaneous fish (Table 10-4). However, the amounts from northern rockfish targeted hauls are likely lower as this includes all rockfish target hauls.

Prohibited species catch in the GOA rockfish fishery is generally low for most species. Catch of prohibited and non-target species generally decreased with implementation of the Central GOA Rockfish Program (Lunsford2013?). Since the 2020 assessment the prohibited species catch observed in 2021 and 2022 increased for Chinook salmon and non-chinook salmon increased, and remained at similar levels for halibut (Table 10-5).

In summary, northern rockfish are most likely to be associated with other rockfish fisheries and the bycatch of non-rockfish species in the northern rockfish fishery are likely low but the only data available is for all rockfish-targeted hauls. Bycatch estimates decreased for the majority of species in the Central GOA following the implementation of the Rockfish Pilot Program. The significant prohibited species that are encountered are Pacific halibut, Chinook and non-Chinook salmon.

Gulf-wide discard rates (percent of the total catch discarded within management categories; Table-1) of northern rockfish are show for 1993–2022. These rates are generally considered to be low and are consistent with other GOA rockfish species. These discard rates are generally similar to those in the GOA for Pacific ocean perch and dusky rockfish. Discard mortality is assumed to be 100% for GOA northern rockfish.

Management Measures

From 1988-1993, the North Pacific Fishery Management Council (NPFMC) managed northern rockfish in the Gulf of Alaska as part of the slope rockfish assemblage. In 1991, the NPFMC divided the slope rockfish assemblage in the Gulf of Alaska into three management subgroups: Pacific ocean perch, shortraker/rougheye rockfish, and a complex of all other species of slope rockfish, including northern rockfish. In 1993, a fourth management subgroup, northern rockfish, was also created. In 2004, rougheye rockfish and shortraker rockfish were also split and managed separately. These subgroups were established to protect Pacific ocean perch, shortraker/rougheye, and northern rockfish (the four most sought-after commercial species in the assemblage) from possible overfishing. Each subgroup is now assigned an individual ABC (acceptable biological catch) and TAC (total allowable catch). Prior to 1991, an ABC and TAC were assigned to the entire assemblage. In the assessments after 1991 and until this year’s assessment, ABC and TAC for each subgroup, including northern rockfish, is apportioned to the three management areas of the Gulf of Alaska (Western, Central, and Eastern) based on a weighted average of the proportion of biomass by area from the three most recent Gulf of Alaska trawl surveys. In this year’s assessment ABC and TAC is apportioned to the three management areas in the Gulf of Alaska with the random effects model developed by the Plan Team survey averaging working group. Northern rockfish are scarce in the eastern Gulf of Alaska, and the ABC apportioned to the Eastern Gulf management area is small. This translates to a TAC that is too difficult to be managed effectively as a directed fishery. Since 1999, the ABC for northern rockfish apportioned to the Eastern Gulf management area is included in the West Yakutat ABC for “other slope rockfish.”

Amendment 41, which took effect in 2000, prohibited trawling east of 140 degrees W. longitude in the Eastern GOA. However, trawling has not occured in this area since 1998. Since most slope rockfish, especially Pacific ocean perch, are caught exclusively with trawl gear, this amendment could have concentrated fishing effort for slope rockfish in the Eastern area in the relatively small area between 140\(^\circ\) and 147\(^\circ\) W longitude that remained open to trawling. This probably does not have a major effect on northern rockfish populations because their abundance in the Eastern area is low.

In November, 2006, NMFS issued a final rule to implement Amendment 68 of the GOA groundfish Fishery Management Plan for 2007 through 2011. This action implemented the Central GOA Rockfish Pilot Program (RPP). The intention of this Program was to enhance resource conservation and improve economic efficiency for harvesters and processors in the rockfish fishery. An additional objective was to spread out the fishery in time and space, allowing for enhanced market conditions for product and reducing the pressure of what was an approximately two-week fishery in July. The primary rockfish management groups in this program are northern rockfish, Pacific ocean perch, and pelagic shelf rockfish. Potential effects of this program on northern rockfish include: 1) Extended fishing season lasting from May 1 – November 15, 2) changes in spatial distribution of fishing effort within the Central GOA, 3) improved at-sea and plant observer coverage for vessels participating in the rockfish fishery, and 4) a higher potential to harvest 100% of the TAC in the Central GOA region. In a comparison of catches in the four years before the RPP to the four years after, it appears that average catches have increased overall (although, this may be due to increased observer coverage) and have spread out spatially in the western and central Gulf (see Figure 10.1 in Hulson2013?). The authors will continue to monitor the benefits and consequences of this action. A summary of key management measures and a time series of catch, ABC and TAC are provided in Table 10-2.

Data

The following table summarizes the data used in the stock assessment model for northern rockfish (* denotes new data for this assessment):

Fishery

Catch

Catch of northern rockfish ranges from 185 t to 17,430 t during 1961-2022. Detailed descriptions of catch are provided in Table 10-?? and Figure 10-3. This is the commercial catch history used in the assessment model. In response to Annual Catch Limits (ACLs) requirements, assessments now document all removals including catch that is not associated with a directed fishery. Estimates of all removals not associated with a directed fishery including research catches are available and are presented in Appendix 10a. In summary, annual research removals have typically been less than 100 t and very little is taken in recreational or Pacific halibut fisheries. These levels likely do not pose a significant risk to the northern rockfish stock in the GOA.

Age and Size Composition

Observers aboard fishing vessels and at onshore processing facilities have provided data on size and age composition of the commercial catch of northern rockfish. Ages were determined from the break-and-burn method (Chilton and Beamish 1982). Length compositions are presented in Table 10-8 and Figure 10-?? and age compositions are presented in Table 10-6 and Figure 10-6; these tables also include associated annual sample sizes and number of hauls sampled for the age and length compositions. The fishery age compositions indicate that stronger than average year-classes occurred around the year 1976 and 1984. The fishery age compositions from 2004 and 2006 also indicate that the 1996-1998 year-classes were strong. There are few younger fish observed in the age compositions for more recent years. The clustering of several large year-classes in each period is most likely due to aging error. Recent fishery length compositions (2003-present) indicate that a large proportion of the northern rockfish catch are found to be larger than 38 cm, which was the previous plus length bin. Length composition data show a slight increase in the size of fish caught, this is well aligned with the lack of younger fish observed in the fishery age compsitions for the same time periods.

Survey

Biomass Estimates from Trawl Surveys

Bottom trawl surveys were conducted in the GOA triennially from 1984–1999 and biennially from since 2001. The surveys provide an index of biomass, size and age composition data, and growth characteristics. The trawl surveys have used a stratified random design to sample fishing stations that cover all areas of the GOA out to a depth of 1,000 m (in some surveys only to 500 m). Generally, attempts have been made through the years to standardize the survey design and the fishing nets used, but there have been some exceptions to this standardization. In particular, much of the survey effort in 1984 and 1987 was by Japanese vessels that used a very different net design and a different survey deisgn than what has been the standard used by U.S. vessels throughout the surveys. To deal with this problem the 1980s survey data have been excluded from this assessment.

Gulf-wide biomass estimates from the VAST model-based index are presented in Table 10-10 and Figure 10-#ref(fig:fvast). The spatial distribution of the catches of northern rockfish in the 2017, 2019, and 2021 surveys are shown in Figure 10-#ref(fig:fcpue). The magnitude of catch varies greatly with several large tows typically occurring in each survey. The precision of some of the biomass estimates has been low and is reflected in the high CVs associated with some survey biomass estimates of northern rockfish that are the result of few very large catches during the survey. In 2001, a single very large survey haul of northern rockfish greatly increased the biomass estimates and resulted in wide confidence bounds. The haul in 2001 was the largest individual catch (14 t) of northern rockfish ever taken during a GOA survey; this tow accounted for 58.7% of total survey catch by mass in that year. In contrast, the 2005 and 2007 survey had several large hauls of northern rockfish in the Central Gulf with similar confidence bounds. Due to the substantial variability observed in the design-based index this assessment is using the VAST model-based index of abundance, though trawl survey biomass from a design-based estimator is also presented per SSC request (Table 10-??).

Age and Size Composition

Ages for northern rockfish were determined from the break-and-burn method (Chilton and Beamish 1982). These age compositions (Table 10-7 and Figure 10-??) indicate that recruitment of northern rockfish is highly variable. The 1990 and 1996 surveys show especially strong year-classes from the period around 1975-77; although they differ as to which specific years were greatest, likely due to age determination errors. The 1993, 1996, and 1999 age compositions also indicate that the 1983-85 year-classes may be stronger than average. Recent age compositions (2005-2011) indicate that the 1996-98 year-classes may also be stronger than average, which is in agreement with recent age compositions obtained from the commercial fishery described above. Trawl surveys provide size composition data for northern rockfish but are not used directly in the current age structured assessment model (Table 10-9 and Figure 10-??). In years with age readings, trawl survey size composition data are multiplied by an age-length key (computed from length-stratified otolith collections) to obtain survey age compositions. Similar to the fishery length compositions discussed above, a large proportion of northern rockfish lengths are greater than the previous plus length bin (38 cm); especially in recent years. However, this issue has been addressed with the increased plus size group in this assessment. Also similar to the fishery age compositions, the proportion of older fish been increasing since the mid to early 2000s with few younger fish observed.

Maturity Data

In previous stock assessments for northern rockfish, age at maturity was based on a logistic curve fit to ovarian samples collected from female northern rockfish in the central GOA in the spring of 1996 (n=75, C. Lunsford pers. comm. July 1997, Heifetz et al. 2009). A study reevaluating maturity of northern rockfish (Chilton 2007 n=157) provides additional information for maturity-at-age. This study collected ovarian samples from female northern rockfish throughout the year in both 2000 and 2001. In a report submitted to the GOA Groundfish Plan Team in September 2010, the two studies were compared and the advantages and disadvantages of the different approaches for studying maturity (histology versus visual inspection) were discussed (Rodgveller2010?). In this year’s assessment, as in the 2020 assessment, we combine the data from both studies to estimate maturity of northern rockfish. Due to the relatively small sample sizes for each study, the close proximity in time for each study (4 years apart compared to the 51 year time series used in this assessment), and the large difference in the age at 50% maturity (12.8 years used in previous assessments compared to 8 years obtained by Chilton 2007), we combine these data and estimate an intermediate maturity-at-age rather than consider time-dependent changes in maturity (Figure 10-??). There could be time-dependent changes in maturity-at-age for northern rockfish, although, additional data would be necessary to evaluate this hypothesis. More recently, Conrath (2019) has reported skip spawning in northern rockfish, the impacts of which are not currently incorporated into the assessment as the spatial and temporal aspects are unknown.

Analytical approach

General Model Structure

The Gulf of Alaska northern rockfish is described as a separable age-structured model assessment is based on a statistical age-structured model with the catch equation and population dynamics model as described in (Fournier1982?) and elsewhere (e.g., (Hilborn1992?); (Schnute1995?), (McAllister1997?)) and was implemented using AD Model Builder software (Fournier et al. 2012). The assessment model is based on a generic rockfish model developed in a workshop held in February 2001 (Courtney et al. 2007) and follows closely the GOA Pacific ocean perch model (Hulson et al. 2021). The northern rockfish model is fit to a time series extending from 1961-2022. As with other rockfish age-structured models, this model does not attempt to fit a stock-recruitment relationship but estimates a mean recruitment, which is adjusted by estimated recruitment deviations for each year. We do this because there does not appear to be an obvious stock-recruitment relationship in the model estimates, and there have been very high recruitments at low stock size (Figure 10-9). The parameters, population dynamics, and equations of the model are shown below.

Description of Alternative Models

A suite of incremental models were run to investigate the effects of removing 1980s survey data from the assessment and increasing the length composition plus group . For clarity the author’s preferred model (m22.1) and the base model (m18.2b; updated with 2022 data) and m22 (base model with corrected survey years) are presented here. The author’s preferred model is the base model with increased length plus groups and uses a VAST model-based index of survey abundance for 1990+. This model generally produces good visual fits to the data and biologically reasonable patterns of recruitment, abundance, and selectivity, and a reasonable retrospective Mohn’s rho value.

Parameters Estimated Outside the Assessment Model

A von Bertalanffy growth curve was fitted, for both sexes combined, to survey size at age data from 1990-2021 using length-stratified methods (Bettoli and Miranda 2001; Quinn1999?). An age to size conversion matrix was then constructed by adding normal error with a standard deviation equal to the survey data for the probability of different sizes for each age class. The length-weight relationship for combined sexes, using the formula \(W = aL^b\), where \(W\) is weight in grams and \(L\) is fork length in mm, \(a\) = 0.0783 and \(b\) = 2.0742. Previous parameters are available from (Heifetz199?); Courtney et al. (1999); and Malecha et al. (2007). The estimated parameters for the growth curve from length-stratified methods are:
\(L_\infty\) = 41.68 cm, \(\kappa\) = 0.16, and \(t_0\) = -0.38.

Weight-at-age was constructed with weight at age data from the same data set as the length at age. Mean weight-at-age is approximated by the equation: \[W_a =W_\infty (1-e^(-\kappa(age-t_0 ) ) )^b.\] The estimated growth parameters from length-stratified methods are:
\(W_\infty\) = 1081 g, \(\kappa\) = 0.17, \(t_0\) = -0.16, and b = 3.04.

Aging error matrices were constructed by assuming that the break-and-burn ages were unbiased but had a given amount of normal error around each age based on between-reader percent agreement tests conducted at the AFSC Age and Growth lab. We fix the variability of recruitment deviations (\(\sigma_r\)) at 1.5 which allows for highly variable recruitment.

Parameters Estimated Inside the Assessment Model

The estimates of natural mortality (M) and catchability (q) are computed with the use of lognormal prior distributions as penalties that are added to the overall objective function in order to constrain parameter estimates to reasonable values and to speed model convergence. Arithmetic means and standard errors (\(\mu, \sigma\)) for the lognormal distributions were provided as inputs to the model. The standard errors for selected model parameters were estimated based on multivariate normal approximation of the covariance matrix. The prior mean for natural mortality of 0.06 is based on the estimate provided by (Heifetz1991?) using the methods of Alverson and Carney (1975). Natural mortality is a difficult parameter to estimate within the model so we assign a “tight” prior CV of 5%. Catchability is a parameter that is unclear for rockfish, so while we assign it a prior mean of 1 (assuming all fish in the area swept are captured and there is no herding of fish from outside the area swept, and that there is no effect of untrawlable grounds), we assign it a less precise CV of 45%. This allows the parameter more freedom than that allowed to natural mortality. These methods are also used in the GOA Pacific ocean perch and GOA dusky rockfish assessments.

The fit to the combined observations of maturity-at-age obtained in the preferred assessment model is shown in Figure 10-15. Parameters for the logistic function describing maturity-at-age estimated conditionally in the model, as well as all other parameters estimated conditionally, were identical to estimating maturity-at-age independently. Estimating maturity-at-age parameters conditionally influences the model only through the evaluation of uncertainty, as the MCMC procedure includes variability in the maturity parameters in conjunction with variability in all other parameters, rather than assuming the maturity parameters are fixed. Thus, estimation of maturity-at-age within the assessment model allows for uncertainty in maturation to be incorporated into uncertainty for key model results (e.g., ABC).

Given that we are using Bayesian estimation, there is no need to implement a recruitment bias-correction algorithm (e.g., Methot2011?).

The numbers of estimated parameters from the model are:

Evaluation of model uncertainty is obtained through a Markov Chain Monte Carlo (MCMC) algorithm (Gelman et al. 1995). The chain length of the MCMC was 10,000,000 and was thinned to one iteration out of every 2,000. We omit the first 1,000,000 iterations to allow for a burn-in period. We use these MCMC methods to provide further evaluation of uncertainty in the results below including 95% credible intervals for some parameters (computed as the 5^th and 95^th percentiles of the MCMC samples).

Results

Model Evaluation

The author’s preferred model is model m22.1. The examined models described here are:

When we present alternative model configurations, our usual criteria for choosing a superior model are: (1) the best overall fit to the data (in terms of negative log-likelihood), (2) biologically reasonable patterns of estimated recruitment, catchabilities, and selectivity, (3) a good visual fit to length and age compositions, and (4) parsimony. We’ve presented results for multiple models because the 2020 and 2022 models differ in either the geospatial model (VAST) timeseries inputs were changed and/or the length plus group was increased. Model m22.1 produces good visual fits to the data, and biologically reasonable patterns of recruitment, abundance, and selectivities. Therefore, the recommended 2022 model is utilizing the new information effectively, and we use it to recommend the 2023 ABC and OFL.

Time Series Results

Key results have been summarized in Tables 10-11 – 10-??. In general, model predictions continue to fit the data well (Figures 10-3, 10-4, 10-10, 10-11, 10-12)

Definitions
Spawning biomass is the biomass estimate of mature females in tons. Total biomass is the biomass estimate of all northern rockfish age-2 and greater in tons. Recruitment is measured as number of age-2 northern rockfish. Fishing mortality is fully-selected F, meaning the mortality at the age the fishery has fully selected the fish.

Biomass and Exploitation Trends
The predicted

The estimates of current population abundance indicate that it is dominated by fish from the 1993 and 1998 year-classes (Table 10-??). Since the early 1990s the total biomass estimated in the model has been decreasing from a high of over 189,000 t in 1992 (Figure 10-8). Similarly, the spawning biomass estimated in the model has also been decreasing since 1998. However, the fit to the VAST model-based survey biomass index fails to capture the apparent increase in northern rockfish abundance indicated by point estimates of the 2005, 2007, 2013, and 2017 trawl surveys (Figure 10-4). Higher survey indices in these years may represent significant abundances of northern rockfish that are not fully accounted for in assessed biomass, but may also simply represent variation in survey catchability.

Spawning biomass estimates have continued to decline since about 2010 (2). Total age-2+ biomass estimates for all model using VAST survey inputs also indicate a steadily decline trend with a peak around 2002.

The estimated selectivity curve for the fishery and survey data suggested a pattern similar to previous assessments for northern rockfish (Figure 10-15). The commercial fishery targets slightly larger and (likely) older fish and the survey should sample a larger range of ages. Ninety-five percent of northern rockfish are selected in the fishery by age 10. The age at 50% selection is 9.1 for the survey and 8.2 for the fishery, age at 50% maturity is estimated at 10.6 years.

Goodman et al. (2002) suggested that stock assessment authors use a “management path” graph as a way to evaluate management and assessment performance over time. In the management path we plot the ratio of fishing mortality to \(F_{OFL}\) (\(F_{35\%}\)) and the estimated spawning biomass relative to \(B_{35\%}\). Harvest control rules based on \(F_{35\%}\) and \(F_{40\%}\) and the tier 3b adjustment are provided for reference. The historical management path for northern rockfish has been above the \(F_{OFL}\) adjusted limit for only a few years in the 1960s. In recent years, northern rockfish have been above \(B_{35\%}\) and below \(F_{35\%}\) (Figure 10-16). The trajectory of fishing mortality has remained below the \(F_{40\%}\) level most of the time and below \(F_{35\%}\) in all years except 1964-76 during the period of intense fishing for Pacific ocean perch. Parameter estimates from this year’s model were similar to the previous northern rockfish assessment (Table 10-11). Selectivity estimates for the fishery and the survey are similar, but with the survey selectivity increasing somewhat more gradually with age. Compared to the maturity at age curve that is estimated, selectivity occurs at slightly younger ages than the age of maturity (Table 10-13 and Figure 10-10).

Recruitment
Recruitment estimates show a high degree of uncertainty, but indicate several large year-classes in the early and late 1970’s, early 1980’s and mid 1990’s (Table 10-14 and 10-15 and Figure 10-8). Recruitment since 2005 has been considerably lower than the 1970–2005 time period. There is no clear trend between recruitment and spawning stock biomass (Figure 10-??). Fits to the fishery and survey age compositions were reasonable with this year’s recommended model (Figures 10-10 and 10-11). Increasing proportions of GOA northern rockfish in the plus age or length groups for both survey and fishery composition indicate a substantial number of individuals are successfully surviving natural and fishing mortality to attain old age and large size.

Retrospective analysis
From the MCMC chains described in the Uncertainty approach section, we summarize the posterior densities of key parameters for the recommended model using histograms (Figure 10-17) and credible intervals (Table 10-12). We also use these posterior distributions to show uncertainty around time series estimates such as total biomass, recruitment, and spawning biomass (Figures 10-6, 10-8, 10-14).

Table 10-12 shows the maximum likelihood estimate (MLE) of key parameters with their corresponding standard deviations derived from the Hessian matrix compared to the standard deviations derived from MCMC methods. The Hessian and MCMC standard deviations are larger for the estimates of q, \(F_{40\%}\), ABC, and female spawning biomass. These larger standard deviations indicate that these parameters are more uncertain than indicated by the standard estimates. However, all estimates fall within the Bayesian credible intervals. The distributions of these parameters are slightly skewed with higher MLE estimates than MCMC medians for current spawning and total biomass and ABC, indicating possibilities of higher biomass estimates (Figure 10-17). Uncertainty estimates in the time series of spawning biomass also result in a skewed distribution towards higher values, particularly at the end of the time series and into the 15 year projected times series (Figure 10-14).

A within-model retrospective analysis of the recommended model was conducted for the last 10 years of the time-series by dropping data one year at a time. The revised Mohn’s “rho” statistic (Hanselman2013?) in female spawning biomass was -0.082, indicating that the model slightly increases the estimate of female spawning biomass in recent years as data is added to the assessment. The retrospective female spawning biomass and the relative difference in female spawning biomass from the model in the terminal year are shown in Figure 10-18 (with 95% credible intervals from MCMC).

Harvest recommendations

Amendment 56 Reference Points

Amendment 56 to the GOA Groundfish Fishery Management Plan defines the “overfishing level” (OFL), the fishing mortality rate used to set OFL (\(F_{OFL}\)), the maximum permissible ABC, and the fishing mortality rate used to set the maximum permissible ABC. The fishing mortality rate used to set ABC (\(F_{ABC}\)) may be less than this maximum permissible level, but not greater. Because reliable estimates of reference points related to maximum sustainable yield (MSY) are currently not available but reliable estimates of reference points related to spawning per recruit are available, Northern rockfish in the GOA are managed under Tier 3 of Amendment 56. Tier 3 uses the following reference points: \(B_{40\%}\), equal to 40% of the equilibrium spawning biomass that would be obtained in the absence of fishing; \(F_{35\%}\),,equal to the fishing mortality rate that reduces the equilibrium level of spawning per recruit to 35% of the level that would be obtained in the absence of fishing; and \(F_{40\%}\), equal to the fishing mortality rate that reduces the equilibrium level of spawning per recruit to 40% of the level that would be obtained in the absence of fishing. Estimation of the \(B_{40\%}\) reference point requires an assumption regarding the equilibrium level of recruitment. In this assessment, it is assumed that the equilibrium level of recruitment is equal to the average of age-2 recruitments between 1979 and 2020. Because of uncertainty in very recent recruitment estimates, we lag 2 years behind model estimates in our projection. Other useful biomass reference points which can be calculated using this assumption are \(B_{100\%}\) and \(B_{35\%}\), defined analogously to \(B_{40\%}\). The 2022 estimates of these reference points are:

\(B_{100\%}\)	\(B_{40\%}\)	\(B_{35\%}\)	\(F_{40\%}\)	\(F_{35\%}\)
82,350	32,940	28,822	0.074	0.061

Specification of OFL and Maximum Permissible ABC

Female spawning biomass for 2022 is estimated at 39,445 t. This is above the \(B_{40\%}\) value of 32,940 t. Under Amendment 56, Tier 3, the maximum permissible fishing mortality for ABC is \(F_{40\%}\) and fishing mortality for OFL is \(F_{35\%}\). Applying these fishing mortality rates for 2022, yields the following ABC and OFL:

ABC	OFL
4,965	5,927

A standard set of projections is required for each stock managed under Tiers 1, 2, or 3 of Amendment 56. This set of projections encompasses seven harvest scenarios designed to satisfy the requirements of Amendment 56, the National Environmental Policy Act, and the Magnuson-Stevens Fishery Conservation and Management Act (MSFCMA).

For each scenario, the projections begin with the vector of 2022 numbers at age as estimated in the assessment. This vector is then projected forward to the beginning of 2023 using the schedules of natural mortality and selectivity described in the assessment and the best available estimate of total (year-end) catch for 2022. In each subsequent year, the fishing mortality rate is prescribed on the basis of the spawning biomass in that year and the respective harvest scenario. In each year, recruitment is drawnfrom an inverse Gaussian distribution whose parameters consist of maximum likelihood estimates determined from recruitments estimated in the assessment. Spawning biomass is computed in each year based on the time of peak spawning and the maturity and weight schedules described in the assessment. Total catch after 2022 is assumed to equal the catch associated with the respective harvest scenario in all years. This projection scheme is run 1,000 times to obtain distributions of possible future stock sizes, fishing mortality rates, and catches.

Five of the seven standard scenarios will be used in an Environmental Assessment prepared in conjunction with the final SAFE. These five scenarios, which are designed to provide a range of harvest alternatives that are likely to bracket the final TAC for 2023, are as follow (“\(max F_{ABC}\)” refers to the maximum permissible value of \(F_{ABC}\) under Amendment 56):

• Scenario 1: In all future years, F is set equal to \(max F_{ABC}\). (Rationale: Historically, TAC has been constrained by ABC, so this scenario provides a likely upper limit on future TACs.)

• Scenario 2: In 2022 and 2023, F is set equal to a constant fraction of \(max F_{ABC}\), where this fraction is equal to the ratio of the realized catches in 2019-2021 to the ABC recommended in the assessment for each of those years. For the remainder of the future years, maximum permissible ABC is used. (Rationale: In many fisheries the ABC is routinely not fully utilized, so assuming an average ratio catch to ABC will yield more realistic projections.)

• Scenario 3: In all future years, F is set equal to 50% of \(max F_{ABC}\). (Rationale: This scenario provides a likely lower bound on \(F_{ABC}\) that still allows future harvest rates to be adjusted downward when stocks fall below reference levels.)

• Scenario 4: In all future years, F is set equal to the 2017-2021 average F. (Rationale: For some stocks, TAC can be well below ABC, and recent average F may provide a better indicator of \(F_{TAC}\) than \(F_{ABC}\).)

• Scenario 5: In all future years, F is set equal to zero. (Rationale: In extreme cases, TAC may be set at a level close to zero.)

Two other scenarios are needed to satisfy the MSFCMA’s requirement to determine whether a stock is currently in an overfished condition or is approaching an overfished condition. These two scenarios are as follows (for Tier 3 stocks, the MSY level is defined as \(B_{35\%}\)):

• Scenario 6: In all future years, F is set equal to \(F_{OFL}\). (Rationale: This scenario determines whether a stock is overfished. If the stock is expected to be 1) above its MSY level in 2022 or 2) above ½ of its MSY level in 2022 and above its MSY level in 2032 under this scenario, then the stock is not overfished.)

• Scenario 7: In 2023 and 2024, F is set equal to max \(F_{ABC}\), and in all subsequent years F is set equal to FOFL. (Rationale: This scenario determines whether a stock is approaching an overfished condition. If the stock is 1) above its MSY level in 2024 or 2) above 1/2 of its MSY level in 2024 and expected to be above its MSY level in 2034 under this scenario, then the stock is not approaching an overfished condition.)

Spawning biomass, fishing mortality, and yield are tabulated for the seven standard projection scenarios (Table 10-15). For projections in Scenario 2 (Author’s F); we use pre-specified catches to increase accuracy of short-term projections in fisheries where the catch is usually less than the ABC. This was suggested to help management with setting preliminary ABCs and OFLs for two-year ahead specifications.

In addition to the seven standard harvest scenarios, Amendments 48/48 to the BSAI and GOA Groundfish Fishery Management Plans require projections of the likely OFL two years into the future. While Scenario 6 gives the best estimate of OFL for 2022, it does not provide the best estimate of OFL for 2023, because the mean 2022 catch under Scenario 6 is predicated on the 2022 catch being equal to the 2022 OFL, whereas the actual 2022 catch will likely be less than the 2022 OFL. The executive summary contains the appropriate one- and two-year ahead projections for both ABC and OFL.

Risk Table and ABC recommendation

The SSC in its December 2018 minutes recommended that all assessment authors use the risk table when determining whether to recommend an ABC lower than the maximum permissible. The following template is used to complete the risk table:

The table is applied by evaluating the severity of four types of considerations that could be used to support a scientific recommendation to reduce the ABC from the maximum permissible. These considerations are stock assessment considerations, population dynamics considerations, environmental/ecosystem considerations, and fishery performance. Examples of the types of concerns that might be relevant include the following:

1. “Assessment considerations—data-inputs: biased ages, skipped surveys, lack of fishery-independent trend data; model fits: poor fits to fits to fishery or survey data, inability to simultaneously fit multiple data inputs; model performance: poor model convergence, multiple minima in the likelihood surface, parameters hitting bounds; estimation uncertainty: poorly-estimated but influential year classes; retrospective bias in biomass estimates.
2. “Population dynamics considerations—decreasing biomass trend, poor recent recruitment, inability of the stock to rebuild, abrupt increase or decrease in stock abundance.
3. “Environmental/ecosystem considerations—adverse trends in environmental/ecosystem indicators, ecosystem model results, decreases in ecosystem productivity, decreases in prey abundance or availability, increases or increases in predator abundance or productivity.
4. “Fishery performance—fishery CPUE is showing a contrasting pattern from the stock biomass trend, unusual spatial pattern of fishing, changes in the percent of TAC taken, changes in the duration of fishery openings.”

Assessment considerations

Level 1. In recent assessments the GOA northern rockfish assessment model has resulted in a negative retrospective pattern, which is interpreted as the model continually increases spawning biomass as new data are added (-0.20 in 2018 and -0.24 in 2020, the current assessment, Figure 10-??). While the assessment fits to composition data from the survey (age) and fishery (age) are generally adequate (Figures 10-?? and 10-??), the fishery length composition (Figure 10-??) are misaligned, though this fit has improved with the increase in the length plus group size. Changing from a design-based model to a VAST-based estimate has made the survey biomass estimates more realistic (less overall fluctuation) though the model continues to fit these data poorly. There is some question as to the efficacy of this trawl survey for developing indices of northern rockfish abundance. The items described here have been an issue for assessing northern rockfish for some time , we scored this category as Level 1, as the level of concern has not changed.

Population dynamics considerations

Level 1. Recruitment since 2005 has been considerably lower than in 1970–2005. There is increasing proportions of GOA northern rockfish in the plus age groups for both survey and fishery age composition indicate a substantial number of individuals are successfully surviving natural and fishing mortality to attain older ages and larger sizes. There is a reduction in body condition in recent years for young rockfish, though how this propagates through time is unclear. Skip spawning has been observed for this species, the spatial and temporal extent which is unknown. However, preliminary investigations that incorporate skip spawning in maturity estimates lead to a reduction is spawning biomass and associated ABC.
For these reasons we have given this risk table factor a level 2 concern for population dynamics considerations though make no recommendation for a reduction in ABC.

Environmental/Ecosystem considerations

Level 1. Environmental mechanisms for changes in survival remain unknown, though changes in water temperature and currents could have effects on prey abundance and success of transition of rockfish from pelagic to demersal stage. Predator effects would likely be more important on larval, post-larval, and small juvenile slope rockfish, but there is insufficient information on these life stages and their predators to inform a conclusion. Additionally, changes in bottom habitat due to natural or anthropogenic causes could alter survival rates by altering available shelter, prey, or other functions. Estimates of structural epifauna habitat (with non-targeted data) have recently been in decline. However, given the continued lack of biological and habitat information for northern rockfish, we scored this category as Level 1, as the level of concern has not changed.

Fishery performance

Level 1. Fishers usually direct their efforts first toward Pacific ocean perch because of its higher value relative to northern rockfish. After the TAC for Pacific ocean perch has been reached and NMFS closes directed fishing for this species, trawlers switch and target northern rockfish. The directed GOA northern rockfish fishery is concentrated on a limited number of highly productive locations. The patterns of fishing and percent of TAC taken have not substantially changed in the last three years, therefore we scored this category as Level 1.

Summary and ABC recommendation

We have ranked three categories as ‘Level 1: No apparent concern’ and one as a ‘Level 2, substantially increased concerns’. The GOA northerm rockfish assessment appears to fit available data well, the 2021 GOA trawl survey was undertaken as planned and data are included in this year’s assessment, and the fishery and environmental considerations appear to be within normal bounds. Because GOA northern rockfish ABC is has not been fully utilized in recent years we are not recommending a reduction in ABC at this time. We anticipate that we will monitor the survey abundance estimates, catch rates, and explore skip spawning more fully in the next assessment.

Area Allocation of Harvests

Apportionment of ABC and OFL among regulatory areas has been based on the random effects model developed by the survey averaging working group. The random effects model was fit to the survey biomass estimates (with associated variance) for the Western, Central, and Eastern GOA. The random effects model estimates a process error parameter (constraining the variability of the modeled estimates among years) and random effects parameters in each year modeled. The fit of the random effects model to survey biomass in each area is shown in Figure 10-??.

In general the random effects model fits the area-specific design-based survey biomass estimates reasonably well. Based on the random effects estimates the area apportionments for GOA northern rockfish are 52.7% for the Western area (up from 37.76% in 2021), 47.3% for the Central area (down from 62.22% in 2018), and 0.02% for the Eastern area (same as 2021). Applying the random effect model apportionments to the recommended ABC for northern rockfish results in 2,614 t for the Western area, 2,350 t for the Central area, and 1 t for the Eastern area for 2023. For management purposes, the small ABC of northern rockfish in the Eastern area is combined with the Other Rockfish complex.

Status Determination

Under the MSFCMA, the Secretary of Commerce is required to report on the status of each U.S. fishery with respect to overfishing. This report involves the answers to three questions: 1) Is the stock being subjected to overfishing? 2) Is the stock currently overfished? 3) Is the stock approaching an overfished condition?

Is the stock being subjected to overfishing? The official catch estimate for the most recent complete year (2021) is 2,376 t. This is less than the 2021 OFL of 0 t. Therefore, the stock is not being subjected to overfishing.

Harvest Scenarios #6 and #7 are intended to permit determination of the status of a stock with respect to its minimum stock size threshold (MSST). Any stock that is below its MSST is defined to be overfished. Any stock that is expected to fall below its MSST in the next two years is defined to be approaching an overfished condition. Harvest Scenarios #6 and #7 are used in these determinations as follows:

Is the stock currently overfished? This depends on the stock’s estimated spawning biomass in 2022:

• 1. If spawning biomass for 2022 is estimated to be below ½ \(B_{35\%}\), the stock is below its MSST.
• 2. If spawning biomass for 2022 is estimated to be above \(B_{35\%}\) the stock is above its MSST.
• 3. If spawning biomass for 2022 is estimated to be above ½ \(B_{35\%}\) but below \(B_{35\%}\), the stock’s status relative to MSST is determined by referring to harvest Scenario #6 (Table 10-15). If the mean spawning biomass for 2034 is below \(B_{35\%}\), the stock is below its MSST. Otherwise, the stock is above its MSST.

Is the stock approaching an overfished condition? This is determined by referring to harvest Scenario #7:

• 1. If the mean spawning biomass for 2024 is below 1/2 \(B_{35\%}\), the stock is approaching an overfished condition.
• 2. If the mean spawning biomass for 2024 is above \(B_{35\%}\), the stock is not approaching an overfished condition.
• 3. If the mean spawning biomass for 2024 is above 1/2 \(B_{35\%}\) but below \(B_{35\%}\), the determination depends on the mean spawning biomass for 2034 If the mean spawning biomass for 2034 is below \(B_{35\%}\), the stock is approaching an overfished condition. Otherwise, the stock is not approaching an overfished condition.

Based on the above criteria and Table 10-15, the stock is not overfished and is not approaching an overfished condition. The fishing mortality that would have produced a catch for last year equal to last year’s OFL is 0.0641.

Ecosystem Considerations

In general, a determination of ecosystem considerations for GOA northern rockfish is hampered by a lack of biological and habitat information. However, a review of the most recent (2021) GOA Ecosystem Status Report did not reveal strong evidence of declining trends in indicators which results in strong concern for northern rockfish. Information regarding the FMP, non-FMP, and prohibited species caught in rockfish target fisheries to help understand ecosystem impacts by the northern fishery (Tables 12-3–12-5).

Ecosystem Effects on the Stock

Prey availability/abundance trends: Unfortunately, there is no information on the food habits of larval or post-larval rockfish to help determine possible relationships between prey availability and year-class strength. Moreover, identification to the species level for field collected larval slope rockfish is difficult. Visual identification is not possible, though genetic techniques allow identification to species level for larval slope rockfish. Some juvenile rockfish found in inshore habitat feed on shrimp, amphipods, and other crustaceans, as well as some mollusk and fish. Adult northern rockfish feed on euphausiids. Euphausiids are also a major item in the diet of walleye pollock. Changes in the abundance of walleye pollock could lead to a corollary change in the availability of euphausiids, which could then impact northern rockfish. Northern rockfish body condition in 2021 improved from 2019, but continued to be below average, a persistent trend since 2013. Limited information on biomass of calanoid copepod and euhausiids (adult prey) in 2022 indicate average to above average availability (Hopcroft2022?; Drummond2022?; Hatch2022?; Fergusson2022?). Shrimp (juvenile prey) CPUEs have been increasing in the Chirikof, Yakutat, and Southeastern regions over 2019 and 2021 AFSC bottom trawl surveys, while they have declined in the Kodiak region (Palsson2021?). We have no information on other juvenile prey species, such as myctophids, squids, hermit crabs, and molluscs.

Predator population trends: There is no indication of increased predation or competition on northern rockfish. Potential predators include Pacific halibut and other large fish. Competitors for zooplankton prey could include Pacific Ocean perch (Hulson et al. 2021), which remains in high abundance, and walleye pollock, which slightly declined in 2022 (Monahan2022?). Predator effects would likely be more important on larval, post-larval, and small juvenile northern rockfish, but information on these life stages and their predators is lacking. However, survival of larvae are thought to be more related to the abundance and timing of prey availability than predation, due to the lack of rockfish as a prey item commonly found in diets.

Changes in physical environment: Changes in structural habitat present a potential concern for northern rockfish. Vertical structure, including sponges, corals, and rocky habitat, is important habitat for northern rockfish and has experienced multi-year decline (with high uncertainty) across the GOA. Observations in 2021 from AFSC’s bottom trawl and observer data of non-target catches (both not designed to sample structural epifauna and associated with high uncertainty) can be used to monitor trends in structural epifauna, although with uncertainty as these surveys/fisheries are not designed to target these species (Whitehouse2021?).
By combining this fishery independent (AFSC survey) and fishery dependent (observer data) datasets, however, we can see a consistent trend rise above potential variability due to potential gear and effort changes (observer data) and the non-targeted sampling of both methods. A VAST model was run for gorgonian corals, pennatulaceans (e.g., sea pens), and sponges integrating and modeling trawls station densities across the Gulf of Alaska (Palsson 2021). The coral abundance index is variable over time but the trend suggests low abundances resulting from the two most recent surveys (2019 and 2021) compared to most index values observed before 2017. The gulf-wide abundance of pennatulaceans shows an increasing trend from 1990 to 2005 and then a variable trend thereafter and a peak in 2017 followed by a decline in 2019. However, the 2021 index value increased from the 2019 value. The trend of sponges shows relative stability until 2015 followed by a continual 7 year decline in the GOA wide index through 2021 to a historic low value. The declines in sponges are driven by trends in western GOA. Sea anemones (not modeled in VAST) declined in Shumangins in 2019 and 2021, and Kodiak experienced a slight decline in 2021.

It is reasonable to expect that the 2022, and predicted 2023 average deeper ocean temperatures, would provide moderate spawning habitat and surface thermal conditions for northern rockfish during a time when they are spawning and growing to a size that promotes over winter survival. Larval abundance of northern rock sole, along with late winter/early spring shelf spawners (e.g., Pacific cod and walleye pollock), associate with cooler winters and enhanced alongshore spring winds. Larval surveys in Shelikof Strait in 2021 observed approximately average abundance of larval rockfish (not identified to species) and age-0 surveys in western GOA (2022) observed above average Pacific cod and walleye pollock larval abundance indicating moderate to above average conditions for northern rock sole. Surface temperatures were below average in the winter, transitioned from below average to above average in the spring (Fergusson2022?), and above average in the summer across the GOA (Seward Line: 12.3°C).
Thermal conditions for adults (banks along shelf edge approximately 75-150m) may have slightly increased to above average in 2022 (longline survey (250m), Siwicke 2022). AFSC bottom trawl survey data revealed a western shift and slight decrease in depth in northern rockfish distribution within the GOA with no relationship to change in temperature.

Euphausiids are also a major item in the diet of walleye pollock. Changes in the abundance of walleye pollock could lead to a corollary change in the availability of euphausiids, which could then impact northern rockfish. The limited information available on temperature and zooplankton indicate average foraging and growing conditions for the zooplanktivorous northern rockfish during 2020. Heat wave conditions occurred during 2020 but were not as severe as 2019 during the summer and fall in the GOA (Barbeaux 2020). Sea surface temperatures were about 1°C above normal in the western GOA and average in the eastern GOA during the 2020 summer (Alaska Center for Climate Assessment & Policy ACCAP, Thoman personal communication). Inside waters of the GOA were slightly more anomalously warm than offshore temperatures (ACCAP). Offshore of Kodiak, waters above the continental shelf along the GAK line remained anomalously warm (0.5°C) at 200-250 m depth in 2020 but cooler than 2019 (Danielson et al. 2020). Along the GOA slope, the AFSC Longline Survey Subsurface Temperature Index indicates above average temperatures at the surface and at depth (250 m) in 2020 relative to the 2005-2019 time series and cooler temperatures in 2020 relative to 2019 (Siwicke personal communication). In the inside waters, Prince William Sound has remained warm since 2014 (Danielson et al. 2020). However, for the inside waters of the eastern GOA, the top 20 m temperatures of Icy Strait in northern southeast Alaska during summer were slightly below average (8.8°C) in 2020 relative to the 23 year time series (1997-2019) (Fergusson2020?). A recent study published in the U.S. West Coast suggests that the warming that occurred during 2014-2016 may have been beneficial for rockfish recruitment (Morgan2019?).

The primary prey of the adult northern rockfish are euphausiids. Warm conditions tend to be associated with zooplankton communities that are dominated by smaller and less lipid rich species in the GOA (Kimmel2019?). There was limited information on zooplankton in 2020. In the inside waters of Icy Strait, northern southeast Alaska, total zooplankton densities were at the 24 year mean and the lipid content of all zooplankton taxa combined examined during 2020 was average for the time series (1997-2020) and similar to 2019 (Fergusson2020?). By taxa, lipid content was above average for the large calanoid copepods, average for hyperiid amphipods, but lower than average for euphausiids, small copepods and gastropods indicating average nutritional quality of the prey field possibly utilized by larval, juvenile, and adult rockfish (Fergusson and Rogers 2020). In the western GOA, the mean biomass of large calanoids and euphausiids averaged over the top 100m south of Seward Alaska during May were about average in 2020 relative to the time series, 1998-2019 (Fergusson2020?).

Fishery Effects on the Ecosystem

Fishery-specific contribution to bycatch of HAPC biota: In the GOA, bottom trawl fisheries for pollock, deepwater flatfish, and Pacific ocean perch account for most of the observed bycatch of coral, while rockfish fisheries account for little of the bycatch of sea anemones, sea whips, and sea pens. The bottom trawl fisheries for Pacific ocean perch and Pacific cod and the pot fishery for Pacific cod account for most of the observed bycatch of sponges (Table 10-4).

Fishery-specific concentration of target catch in space and time relative to predator needs in space and time (if known) and relative to spawning components: The directed slope rockfish trawl fishery that begins in July is concentrated in known areas of abundance and typically lasts only a few weeks. The annual exploitation rates on rockfish are thought to be quite low. Insemination is likely in the fall or winter, and parturition is likely mostly in the spring. While reproductive activities are probably not directly affected by the commercial fishery, there is evidence of skip spawning often caused by an a lack of fertilization (Conrath 2019). If fishing were to reduce the population substantially or cause significant localized depletion it would be possible to increase the amount of observed skip spawning within the stock.

Fishery-specific effects on amount of large size target fish: No evidence for targeting large fish.

Fishery contribution to discards and offal production: Fishery discard rates of northern rockfish during 2009-2020 have been 1.5 – 5.0%.

Fishery-specific effects on age-at-maturity and fecundity of the target fishery: Unknown.

Fishery-specific effects on EFH living and non-living substrate: Unknown, but the heavy-duty “rockhopper” trawl gear commonly used in the fishery can disturb seafloor habitat. Table 10-4 shows the estimated bycatch of living structure such as benthic urochordates, corals, sponges, sea pens, and sea anemones by the GOA rockfish fisheries. The average bycatch of corals/bryozoans and sponges by rockfish fisheries are a large proportion of the catch of those species taken by all Gulfwide fisheries.

Data Gaps and Research Priorities

Life history and habitat utilization

There is little information on larval, post-larval, or early life history stages of northern rockfish. Habitat requirements for larval, post-larval, and early stages are mostly unknown. Habitat requirements for later stage juvenile and adult fish are anecdotal or conjectural. Research needs to be done on the bottom habitat of the major fishing grounds, on what HAPC biota are found on these grounds, and on what impact bottom trawling may have on these biota. Given the substantial influence of maturity-at-age on management quantities (i.e., ABC) and observations of skip spawning (Conrath 2019) we strongly suggest that continued research be devoted to collecting maturity-at-age data for northern and other Gulf of Alaska rockfish. A proposal is currently in the process of being developed that would collect a larger sample size for northern rockfish and compare maturity at age estimates to previous studies. If funded, additional data collected as part of this study would be used to investigate possible time-dependent maturity.

Assessment Data

The highly variable design-based biomass estimates for northern rockfish from bottom trawl survey suggest that the stratified random design of the surveys does a relatively poor job of assessing stock condition of northern rockfish and that a different survey approach may be needed to reduce the variability in biomass estimates. In particular, the last CIE review report recommended that assumptions about extending area-swept estimates of biomass in trawlable versus untrawlable grounds may impact catchability assumptions. The AFSC is currently undertaking a study on habitat classifications so that assumptions about catchability, in particular, time-dependent changes in catchability, can be more rigorously established. To address some of these issues the design-based index has been replaced with a model-based survey biomass index generated by a Vector Autoregressive Spatio-Temporal (VAST) model. The benefits of the VAST model-based approach to survey index standardization are that as a delta-model it partitions the likelihood of trawl survey observations between encounter probability and positive catch rate components, and accounts for spatial and spatio-temporal correlations in survey catch rates. However, this model could benefit from continued examination of appropriate parameterization for northern rockfish which are found in highly “patchy” distributions. Given the high precision of VAST outputs it may prove valuable to the incorporating of an error inflation parameter to increase the variance in VAST models and explore the effect low survey model variance has on resulting assessment outputs.

References

Ackley, D.R. and Heifetz, J. (2001) Fishing practices under maximum retainable bycatch rates in Alaska’s groundfish fisheries. Alaska Fish. Res. Bull. 8, 22–44.

Allen, M.J. and Smith, G.B. (1988) Atlas and zoogeography of common fishes in the Bering Sea and northeastern Pacific.

Alverson, D.L. and Carney, M.J. (1975) A graphic review of the growth and decay of population cohorts. ICES Journal of Marine Science 36, 133–143.

Barbeaux, S. (2020) Fall 2020 marine heatwave. In: Ecosystem Status Report 2020: Gulf of Alaska, Stock Assessment and Fishery Evaluation Report. North Pacific Fishery Management Council, Anchorage, AK.

Berkeley, S.A., Hixon, M.A., Larson, R.J. and Love, M.S. (2004) Fisheries sustainability via protection of age structure and spatial distribution of fish populations. Fisheries 29, 23–32.

Bettoli, P.W. and Miranda, L.E. (2001) Cautionary note about estimating mean length at age with subsampled data. North American Journal of Fisheries Management 21, 425–428.

Beyer, S.G., Sogard, S.M., Harvey, C.J. and Field, J.C. (2015) Variability in rockfish (sebastes spp.) Fecundity: Species contrasts, maternal size effects, and spatial differences. Environmental Biology of Fishes 98, 81–100.

Brodeur, R.D. (2001) Habitat-specific distribution of Pacific ocean perch (Sebastes alutus) in Pribilof Canyon, Bering Sea. Continental Shelf Research 21, 207–224.

Bruin, J.-P. de, Gosden, R.G., Finch, C.E. and Leaman, B.M. (2004) Ovarian aging in two species of long-lived rockfish, Sebastes aleutianus and S. alutus. Biology of Reproduction 71, 1036–1042.

Carlson, H.R. and Straty, R.R. (1981) Habitat and nursery grounds of Pacific rockfish, Sebastes spp., in rocky, coastal areas of southeastern Alaska. Marine Fisheries Review 43.

Chilton, D.E. and Beamish, R.J. (1982) Age determination methods for fishes studied by the groundfish program at the Pacific Biological Station. Department of Fisheries and Oceans Ottawa, Canada.

Chilton, E. (2007) Maturity of female Northern Rockfish Sebastes polyspinis in the central Gulf of Alaska. Alaska Fish. Res. Bull 12, 264–269.

Clausen, D.M. and Heifetz, J. (2002) The northern rockfish, Sebastes polyspinis, in Alaska: commercial fishery, distribution, and biology. Marine Fisheries Review 64, 1–28.

Conrath, C.L. (2019) Reproductive potential of light dusky rockfish (Sebastes variabilis) and northern rockfish (S. polyspinis) in the Gulf of Alaska. Fishery Bulletin 117, 140–151.

Courtney, D.L., Heifetz, J., Sigler, M.F. and Clausen, D.M. (1999) An age structured model of northern rockfish, Sebastes polyspinis, recruitment and biomass in the Gulf of Alaska. Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska as projected for 2000, 361–404.

Courtney, D.L., Ianelli, J.N., Hanselman, D. and Heifetz, J. (2007) Extending statistical age-structured assessment approaches to Gulf of Alaska rockfish (Sebastes spp.). In: Biology, Assessment, and Management of North Pacific Rockfishes. (eds J. Heifetz, J. DiCosimo, A.J. Gharrett, M.S. Love, O’Connell V.M and R.D. Stanley). Alaska sea Grant, University of Alaska Fairbanks, pp 429–449.

Danielson, S.L., Hill, D.F., Hedstrom, K.S., Beamer, J. and Curchitser, E. (2020) Demonstrating a High-Resolution Gulf of Alaska Ocean Circulation Model Forced Across the Coastal Interface by High-Resolution Terrestrial Hydrological Models. Journal of Geophysical Research: Oceans 125, e2019JC015724.

Du Preez, C. and Tunnicliffe, V. (2011) Shortspine thornyhead and rockfish (Scorpaenidae) distribution in response to substratum, biogenic structures and trawling. Marine Ecology Progress Series 425, 217–231.

Fournier, D.A., Skaug, H.J., Ancheta, J., et al. (2012) AD Model Builder: using automatic differentiation for statistical inference of highly parameterized complex nonlinear models. Optimization Methods and Software 27, 233–249.

Gelman, A., Carlin, J.B., Stern, H.S. and Rubin, D.B. (1995) Bayesian data analysis. Chapman; Hall/CRC.

Gharrett, A., Gray, A., Clausen, D. and Heifetz, J. (2003) Preliminary study of the population structure in Alaskan northern rockfish, Sebastes polyspinis, based on microsatellite and mtDNA variation. Fisheries Division, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, Juneau AK 99801.

Gharrett, A.J., Riley, R.J. and Spencer, P.D. (2012) Genetic analysis reveals restricted dispersal of northern rockfish along the continental margin of the Bering Sea and Aleutian Islands. Transactions of the American Fisheries Society 141, 370–382.

Goodman, D., Mangel, M., Parkes, G., Quinn II, T.J., Restrepo, V., Smith, T. and Stokes, K. (2002) Scientific Review of the Harvest Strategy Currently Used in the BSAI and GOA Groundfish Fishery Management Plans. Draft report.North Pacific Fishery Management Council, Anchorage, AK.

Hannah, R.W. and Parker, S.J. (2007) Age-modulated variation in reproductive development of female Pacific Ocean Perch (Sebastes alutus) in waters off Oregon. In: Biology, assessment, and management of North Pacific rockfishes. (eds J. Heifetz, J. DiCosimo, A.J. Gharrett, M.S. Love, O’Connell V.M and R.D. Stanley). Alaska sea Grant, University of Alaska Fairbanks, pp 1–20.

Heifetz, J. (2002) Coral in Alaska: distribution, abundance, and species associations. Hydrobiologia 471, 19–28.

Hulson, P., Williams, B., Fissel, B., Ferriss, M., BE amd Hall, Yasumiishi, E. and Jones, D. (2021) Assessment of the Pacific ocean perch stock in the Gulf of Alaska. In: Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska. North Pacific Fishery Management Council, Anchorage, AK.

Kamin, L.M., Palof, K.J., Heifetz, J. and Gharrett, A.J. (2013) Interannual and spatial variation in the population genetic composition of young-of-the-year Pacific ocean perch (Sebastes alutus) in the Gulf of Alaska. Fisheries Oceanography 23, 1–17.

Krieger, K. (1993) Distribution and abundance of rockfish determined from a submersible and by bottom trawling. Fish. Bull. 91, 87–96.

Krieger, K.J. and Wing, B.L. (2002) Megafauna associations with deepwater corals (Primnoa spp.) in the Gulf of Alaska. Hydrobiologia 471, 83–90.

Laman, E.A., Kotwicki, S. and Rooper, C.N. (2015) Correlating environmental and biogenic factors with abundance and distribution of Pacific ocean perch (Sebastes alutus) in the Aleutian Islands, Alaska. Fishery Bulletin 113.

Leaman, B.M. (1991) Reproductive styles and life history variables relative to exploitation and management of Sebastes stocks. Environmental Biology of Fishes 30, 253–271.

Malecha, P., Hanselman, D. and Heifetz, J. (2007) Growth and mortality of rockfishes (Scorpaenidae) from Alaska waters.

Rodgveller, C.J., Lunsford, C.R. and Fujioka, J.T. (2012) Effects of maternal age and size on embryonic energy reserves, developmental timing, and fecundity in quillback rockfish (Sebastes maliger). Fishery Bulletin 110, 36–45.

Withler, R., Beacham, T., Schulze, A., Richards, L. and Miller, K. (2001) Co-existing populations of Pacific ocean perch, Sebastes alutus, in Queen Charlotte Sound, British Columbia. Marine Biology 139, 1–12.

Yang, M.-S. (1996) Diets of the important groundfishes in the Aleutian Islands in summer 1991. 105 pp. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-60.

Yang, M.-S. (1993) Food habits of the commercially important groundfishes in the Gulf of Alaska in 1990. 150 pp. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-22.

Yang, M.-S. (2000) Food habits of the commercially important groundfishes in the Gulf of Alaska in 1990, 1993, and 1996. 174 pp. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-112.

Tables

Figures

Figure 10.1. Comparisons of observed and predicted GOA northern rockfish trawl survey biomass for different model variants. Error bars are approximate asymptotic 95% confidence intervals of model error. for different model variants.

Figure 10.2. Comparisons of spawning and total biomass for different model variants.

Figure 10.3. Estimated and observed long-term and recent commercial catch of GOA northern rockfish in the Gulf of Alaska.

Figure 10.4. Observed (geostatistical model-based estimates) and predicted GOA northern rockfish trawl survey biomass based on the 2022 recommended model. Error bars are approximate asymptotic 95% confidence intervals of model error.

Figure 10.5. Spatial distribution of northern rockfish in the Gulf of Alaska during the 2017, 2019, and 2021 NMFS trawl surveys.

Figure 10.6. Model estimated total biomass and spawning biomass with 95% credible intervals determined by MCMC (shaded) for Gulf of Alaska northern rockfish.

Figure 10.7. Gulf of Alaska northern rockfish catch/age 2+ biomass ratio with approximate 95% confidence intervals. Observed catch values were used for 1990-2022, the 2022 catch values were estimated using an expansion factor. The horizontal dashed line is the mean value for the entire dataset.

Figure 10.8. Estimates of age-4 recruitment with 95% credible intervals for GOA northern rockfish.

Figure 10.9. Scatterplot of spawner-recruit estimates for the GOA northern rockfish author’s recommended model.

Figure 10.10. Fishery age compositions for GOA northern rockfish. Observed values are bars, lines are the predicted lengths from author’s recommended model.

Figure 10.11. Survey age compositions for GOA northern rockfish. Observed values are bars, lines are the predicted lengths from author’s recommended model.

Figure 10.12. Fishery length compositions for GOA northern rockfish. Observed values are bars, lines are the predicted lengths from author’s recommended model.

Figure 10.13. Survey length compositions (not used in model) for GOA northern rockfish.

Figure 10.14. Median northern rockfish spawning stock biomass from MCMC simulations with Bayesian credible intervals including projections for 2023-2037, when managing under Scenario 2. Assumes the same average yield ratio forward in time. Dotted horizontal line is \(B_{40\%}\) and solid horizontal line is \(B_{35\%}\) based on recruitments from 1977-2018. Each shade is 5% of the posterior distribution.

Figure 10.15. Estimated maturity, fishery and survey selectivities for GOA northern rockfish from the 2022 model.

Figure 10.16. Time series of northern rockfish estimated spawning biomass (SSB) relative to B_(35%) and fishing mortality (F) relative to F_(35%) for author recommended model.

Figure 10.17. Histograms of estimated posterior distributions for key parameters derived (or estimated, in the case of q) from the MCMC for GOA northern rockfish. Vertical black lines represent the maximum likelihood estimate for comparison with the MCMC results.

Figure 10.18. Retrospective peels of estimated female spawning biomass and total biomass for the past 10 years from the recommended model with 95% credible intervals derived from MCMC. Mohn’s rho value for spawning biomass is -0.123 and -0.115 for total biomass.

Appendix 10a. Supplemental catch data

In order to comply with the Annual Catch Limit (ACL) requirements, a dataset has been generated to help estimate total catch and removals from NMFS stocks in Alaska. This dataset estimates total removals that occur during non-directed groundfish fishing activities. This includes removals incurred during research, subsistence, personal use, recreational, and exempted fishing permit activities, but does not include removals taken in fisheries other than those managed under the groundfish FMP. These estimates represent additional sources of removals to the existing Catch Accounting System estimates. For Gulf of Alaska (GOA) northern rockfish, these estimates can be compared to the research removals reported in previous assessments (Heifetz et al. 2009; Table 10 A-1). Northern rockfish research removals are minimal relative to the fishery catch and compared to the research removals of other species. The majority of research removals are taken by the Alaska Fisheries Science Center’s (AFSC) biennial bottom trawl survey which is the primary research survey used for assessing the population status of northern rockfish in the GOA. Other research activities that harvest northern rockfish include longline surveys by the International Pacific Halibut Commission and the AFSC and the State of Alaska’s trawl surveys. Recreational harvest of northern rockfish rarely occurs. Total removals from activities other than a directed fishery have been near 10 t for 2010 – 2017. The 2017 other removals is <1% of the 2018 recommended ABC of 4,529 t and represents a very low risk to the northern rockfish stock. Research harvests from trawl in recent years are higher in odd years due to the biennial cycle of the AFSC bottom trawl survey in the GOA and have been less than 10 t except in 2013 when 18 t were removed. These removals do not pose a significant risk to the northern rockfish stock in the GOA.

References

Heifetz, J., D. Hanselman, J. N. Ianelli, S. K. Shotwell, and C. Tribuzio. 2009. Gulf of Alaska northern rockfish. In Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska as projected for 2010. North Pacific Fishery Management Council, 605 W 4th Ave, Suite 306 Anchorage, AK 99501. pp. 817-874.

Appendix 10b: VAST model-based abundance

Background

Model-based abundance indices have a long history of development in fisheries (Maunder and Punt 2004). We here use a delta-model that uses two linear predictors (and associated link functions) to model the probability of encounter and the expected distribution of catches (in biomass or numbers, depending upon the specific stock) given an encounter (Lo et al. 1992; Stefánsson 1996).
Previous research has used spatial strata (either based on strata used in spatially stratified design, or post-stratification) to approximate spatial variation (Helser et al. 2004), although recent research suggests that accounting for spatial heterogeneity within a single stratum using spatially correlated residuals and habitat covariates can improve precision for the wrestling index (Shelton et al. 2014).
Model-based indices have been used by the Pacific Fisheries Management Council to account for intra-class correlations among hauls from a single contract vessel since approximately 2004 (Helser et al. 2004).
Specific methods evolved over time to account for strata with few samples (Thorson and Ward 2013), and eventually to improve precision based on spatial correlations (Thorson et al. 2015) using what became the Vector Autoregressive Spatio-temporal (VAST) model (Thorson and Barnett 2017).

The performance of VAST has been evaluated previously using a variety of designs.
Research has showed improved performance estimating relative abundance compared with spatially-stratified index standardization models (Grüss and Thorson 2019; Thorson et al. 2015), while other simulation studies have shown unbiased estimates of abundance trends (Johnson et al. 2019).
Brodie et al. (2020) showed improved performance in estimating index scale given simulated data relative to generalized additive and machine learning models.
Using real-world case studies, Cao et al. (2017) showed how random variation in the placement of tows relative to high-quality habitat could be “controlled for” using a spatio-temporal framework, and OLeary et al. (2020) showed how combining surveys from the eastern and northern Bering Sea within a spatio-temporal framework could assimilate spatially unbalanced sampling in those regions. Other characteristics of model performance have also been simulation-tested although these results are not discussed further here.

Settings used in 2021

The software versions of dependent programs used to generate VAST estimates were:

R (>=4.1.0), INLA (21.02.23), TMB (1.7.18), TMBhelper (1.3.0), VAST (3.6.1), FishStatsUtils (2.8.0)

We used a Poisson-link delta-model (Thorson 2018) involving two linear predictors, and a gamma or lognormal distribution to model positive catch rates. We extrapolated catch density using 3705 m (2 nmi) X 3705 m (2 nmi) extrapolation-grid cells; this results in 36,690 extrapolation-grid cells for the eastern Bering Sea, 15,079 in the northern Bering Sea and 26,510 for the Gulf of Alaska (some Gulf of Alaska analyses eliminated the deepest stratum with depths >700 m because of sparse observations, resulting in a 22,604-cell extrapolation grid). We used bilinear interpolation to interpolate densities from 750 “knots” to these extrapolation-grid cells (i.e, using fine_scale=TRUE feature); knots were approximately evenly distributed over space, in proportion to the dimensions of the extrapolation grid. We estimated geometric anisotropy (how spatial autocorrelation declines with differing rates over distance in some cardinal directions than others), and included a spatial and spatio-temporal term for both linear predictors. To facilitate interpolation of density between unsampled years, we specified that the spatio-temporal fields were structured over time as an AR(1) process (where the magnitude of autocorrelation was estimated as a fixed effect for each linear predictor). However, we did not include any temporal correlation for intercepts, which we treated as fixed effects for each linear predictor and year. Finally, we used epsilon bias-correction to correct for retransformation bias (Thorson and Kristensen 2016).

Diagnostics

We checked model fits for evidence of non-convergence by confirming that (1) the derivative of the marginal likelihood with respect to each fixed effect was sufficiently small and (2) that the Hessian matrix was positive definite.
We then checked for evidence of model fit by computing Dunn-Smyth randomized quantile residuals (Dunn and Smyth 1996) and visualizing these using a quantile-quantile plot within the DHARMa R package.
We also evaluated the distribution of these residuals over space in each year, and inspected them for evidence of residual spatio-temporal patterns.

References

Brodie, S.J., Thorson, J.T., Carroll, G., et al. (2020) Trade-offs in covariate selection for species distribution models: A methodological comparison. Ecography 43, 11–24.

Cao, J., Thorson, J., Richards, A. and Chen, Y. (2017) Geostatistical index standardization improves the performance of stock assessment model: An application to northern shrimp in the Gulf of Maine. Canadian Journal of Fisheries and Aquatic Sciences.

Dunn, K.P., and Smyth, G.K. 1996. Randomized quantile residuals. Journal of Computational and Graphical Statistics 5, 1-10.

Grüss, A. and Thorson, J.T. (2019) Developing spatio-temporal models using multiple data types for evaluating population trends and habitat usage. ICES Journal of Marine Science 76, 1748–1761.

Helser, T.E., Punt, A.E. and Methot, R.D. (2004) A generalized linear mixed model analysis of a multi-vessel fishery resource survey. Fisheries Research 70, 251–264.

Johnson, K.F., Thorson, J.T. and Punt, A.E. (2019) Investigating the value of including depth during spatiotemporal index standardization. Fisheries Research 216, 126–137.

Lo, N.C.-h., Jacobson, L.D. and Squire, J.L. (1992) Indices of relative abundance from fish spotter data based on delta-lognornial models. Canadian Journal of Fisheries and Aquatic Sciences 49, 2515–2526.

Maunder, M.N. and Punt, A.E. (2004) Standardizing catch and effort data: A review of recent approaches. Fisheries research 70, 141–159.

O’Leary, C.A., Thorson, J.T., Ianelli, J.N. and Kotwicki, S. Adapting to climate-driven distribution shifts using model-based indices and age composition from multiple surveys in the walleye pollock (gadus chalcogrammus) stock assessment. Fisheries Oceanography.

Shelton, A.O., Thorson, J.T., Ward, E.J. and Feist, B.E. (2014) Spatial semiparametric models improve estimates of species abundance and distribution. Canadian Journal of Fisheries and Aquatic Sciences 71, 1655–1666.

Stefánsson, G. (1996) Analysis of groundfish survey abundance data: Combining the GLM and delta approaches. ICES journal of Marine Science 53, 577–588.

Thorson, J.T. (2018) Three problems with the conventional delta-model for biomass sampling data, and a computationally efficient alternative. Canadian Journal of Fisheries and Aquatic Sciences 75, 1369–1382.

Thorson, J.T. and Barnett, L.A. (2017) Comparing estimates of abundance trends and distribution shifts using single-and multispecies models of fishes and biogenic habitat. ICES Journal of Marine Science 74, 1311–1321.

Thorson, J.T. and Kristensen, K. (2016) Implementing a generic method for bias correction in statistical models using random effects, with spatial and population dynamics examples. Fisheries Research 175, 66–74.

Thorson, J.T., Shelton, A.O., Ward, E.J. and Skaug, H.J. (2015) Geostatistical delta-generalized linear mixed models improve precision for estimated abundance indices for West Coast groundfishes. ICES Journal of Marine Science 72, 1297–1310.

Thorson, J.T. and Ward, E.J. (2013) Accounting for space–time interactions in index standardization models. Fisheries Research 147, 426–433.